Abstract: To extract complex objects from massive disordered point-cloud data, the distribution range of objects is delineated to extract data of the objects by human-computer interaction in three-dimensional visualization environment. Plane view in three-dimensional is used to express the location and distribution of objects in point-cloud visually. Outsourcing polygons dividing objects are made  artificially in plane view by means of 3D interactive technology. Objects are extracted from original point-cloud data according to the outsourcing polygons, and it is simplified by the effectivegrid method. The method is well applied to extract objects from site point-cloud data with 410 millions points.

Key words: disordered point-cloud, 3D visulization, interactive segmentation, effective grid method

摘要： 针对海量无序点云数据中的复杂对象提取问题，利用三维可视化环境中的人机交互功能，划定对象分布范围以提取对象数据。采用三维平面视图直观表达点云中对象的位置及分布；借助三维交互技术，在平面视图中人工划分对象外包多边形；根据外包多边形从原始点云数据中提取对象点云，使用有效网格法进行化简。以4.1亿个数据点的场站点云为例进行应用，取得良好效果。

关键词: 无序点云, 三维可视化, 交互分割, 有效网格法